Nonempirical Calculations on the Conformation ... - ACS Publications

using the V.C.U. IBM 370/145 for the computations de- scribed above. P.V.A. also thanks the Texaco Corp. for a graduate fellowship and Professor Ray ...
1 downloads 0 Views 513KB Size
476

Finally we note that the CNDO/2-NO-CI parametric scheme does a pretty good job of estimating rotational barriers in small molecules. The INDO-NO-CI results are less good and the C N D O / S parameters are not useful for calculating rotational barriers (Table VI). In cases where the S C F process very nearly produces the dominant configuration, the CND0/2-NO-CI' will offer little improvement, but where alternate low-energy configurations are available (delocalization) the natural orbital C I should offer improved electronic structure interpretations.

Acknowledgment. W e wish to thank Dr. Richard Grove, Director of the V.C.U. Computer Center, and his Systems Programmer, Mr. David Brydon, for their cooperation in using the V.C.U. IBM 370/145 for the computations described above. P.V.A. also thanks the Texaco Corp. for a graduate fellowship and Professor Ray Ottenbrite for many discussions and research training. Finally we thank the referees for lively discussion and several very helpful and constructive suggestions, particularly in distinguishing between MO-CI and VB-CI techniques. References and Notes (1)(a) Virginia Commonwealth University; (b) Texaco Graduate ,Fellow, 1971-1973;(c) University of Virginia. (2)( a ) E. Huckel, Z. Phys., 70, 204 (1931);(b) R. S.Mulliken, Phys. Rev., 32, 186 (1928). (3)C. C. J. Roothaan, Rev. Mod. Phys., 23, 69 (1951). (4)R. G. Parr, "Quantum Theory of Molecular Electronic Structure," W. A. Benjamin, New York, N.Y., 1964. (5) W. Heitier and F. London, 2.Phys., 44,455 (1927). (6) L. Pauling, J. Chem. Phys., 1, 280 (1933). (7)E. S. Gould, "Mechanism and Structure in Organic Chemistry," Holt, Rinehart and Winston, New York, N.Y., 1959. (8)R. 8. Woodward and R:Hoffmann, Angew. Chem.. lnt. Ed. Engl., 8, 781 (1969). (9)S.F. Boys and G. B. Cook, Rev. Mod. Phys., 32,285 (1960). (10)I. L. Cooper and R. McWeeny, J. Chem. Phys., 45,226 (1966).

H. Shull, lnt. J. Quantum Chem., 3,523 (1969). C. M. Reeves, Commun. ACM (Ass. Comput. Mach.),9,276 (1966). J. F. Harrison and L. C. Ailen, J. Mol. Spectrosc., 29, 432 (1969). M. Rubenstein and I. Shavitt, J. Chem. Phys., 51,2014 (1969). W. J. Hunt, P. J. Hay, and W. A. Goddard, J. Chem. Phys., 57, 738

(1972). C. Peterson and G. V. Pfeiffer, Theor. Chim. Acta, 26,321 (1972). R. G. A. Maclagan and G. W. Schnuelle, J. Chem. Phys., 55, 5431

(1971). P. 0. Lowdin, Phys. Rev., 97, 1474 (1955). W. Kutzelnigg, Theor. Chim. Acta, 1, 327,343 (1963). C. F. Bender and E. R. Davidson. Phys. Rev., 183,23 (1969).

H. F. Schaeffer, "The Electronic Structure of Atoms and Molecules." Addison-Wesley, Reading, Mass., 1972. F. L. Pilar, "Elementary Quantum Chemistry." McGraw-Hill, New York, N.Y.. 1968. 370. (23) Rumerl'kachr. Ges. Wss. Goftingen, Math.-Phys. Kl.,Fachgruppe 3,

b.337 (1932).

(24)J. A. Pople and D. L. Beveridge, "Approximate Molecular Orbital Theory," McGraw-Hill, New York, N.Y., 1970. (25)S. Green, J. Chem. Phys., 54,827 (1971). (26)J. C. Slater, "Quantum Theory of Molecules and Solids," Vol. 1. Electronic Structure of Molecules, McGraw-Hill, New York, N.Y., 1963,Appendix 4. (27)N. C. Baird and M. J. Dewar, J. Chem. Phys., 50, 1262 (1969). (28)0.Sinanoglu and H. 0. Pamuk. J. Amer. Chem. Soc., 95,5435 (1973). (29)J. L. Whitten, J. Chem. Phys., 44,359 (1966). (30)E. Scarzafava and L. C. Allen, J. Amer. Chem. Soc., 93, 311 (1971). (31)C. Trindle and D. D. Shillady, J. Amer. Chem. Soc., 95,703 (1973). (32)J. A. Kroll and D. D. Shillady, J. Amer. Chem. SOC.,95, 1422 (1973). (33)J. Del Bene and H. H. Jaffe, J. Chem. Phys., 48, 1807,4050 (1968); ibid., 58,2238 (1973). (34)R. S.Mulliken, J. Chem. Phys., 23, 1833,1841 (1955). (35)K. B. Wiberg, Tetrahedron, 24, 1083 (1968). (36)C. Trindle. J. Amer. CheE. SOC.,91,219 (1969). (37)C. Trindle and 0. Sinanogiu. J. Amer. Chem. SOC., 91,853 (1969). (38)G. Doggett, J. Chem. SOC.A, 229 (1969). (39)D. D:Shillady, F. P. Biiiingsiey, and J. E. Bioor, Theor. Chim. Acta, 21, 1 (1971). (40)H. Shull and P. 0. Lowdin. J. Chem. Phys., 23, 1565 (1955). (41)A. Graovac, I. Gutman, M. Pandic, and N. Trinajstic, J. Amer. Chem. Soc., 95,6267 (1973). (42)M. H. Whangbo and I. Lee, J. Amer. Chem. SOC., 93,2330(1971). (43)P. F. Alewood, P. M. Kazmaier, and A. Rauk. J. Amer. Chem. SOC.,95, 5466 (1973);particularly their footnotes 46 and 54. (44)N. Mataga and K. Nishimoto, 2.Phys. Chem. (Frankfurt a m Main). 13, 140 (1957). (45)D. T. Clarke and J. W. Emsley, Mol. Phys., 12,365 (1967).

Nonempirical Calculations on the Conformation and Hyperfine Structure of the Nitroxide and Ketyl Groups. Consequences of Out-of-Plane Bending on Hyperfine Interactions Yves Ellinger,la.bRobert Subra,*lb Andre Rassat,lbJCr6me Douady,lc and Gaston BerthierIc Contribution from the Laboratoire de Chimie Organique Physique, Dkpartement de Recherche Fondamentale, Centre d'Etudes Nucli?aires de Grenoble, and C.I.S.I.,Cedex 85, F 38041 Grenoble, France, and Laboratoire de Chimie, Ecole Normale Supgrieure de Jeunes Filles, F 921 20 Montrouge, France. Received January 28, 1974

Abstract: Nonempirical calculations of the ground-state energy and hyperfine coupling constants of the isoelectronic H2NO and H2CO- radicals are performed using the spin-restricted SCF method and first-order perturbation theory. It is shown that the nitroxide functional group does not possess a well-defined intrinsic geometry, in agreement with experiment. The equilibrium geometry found for the ketyl is nonplanar, with an out-of-plane angle of 24-27' and close to that of the excited (n-n*) neutral molecule. The nitrogen and carbon-13 splittings a r e positive and increase with the out-of-plane torsion a t the radical site.

Nitroxides a r e one of the most remarkable series of free the interest of which has grown in recent years as a consequence of their extensive use for biological studies. Esr measurements in spin-label experimentsZbare generally interpreted on the assumption of a planar radical site in Journal of the American Chemical Society

/

97:3

/

spite of the fact that the nitroxide group is known to exist in both planar3 ( a = Oo) and bent4 ( a # 0') geometries. On the other hand, the question of the geometry around the radical carbon in the isoelectronic ketyl series has recently been the object of both experimental5 and theoreticalh in-

February 5, I975

411 Table I. Total Energies of HlNO (au)

vestigations.

/

U

Basis I

0 30

0

Basis I1

17 0 26”22‘

Basis I11

-130,027677 -130,026646 -130,125392 -130.125494 -126.59425 - 126.59411 -126.5973a

-130.035687 - 130.034443 -130.140818 -130.140843

UHF calculation.

This prompted us to investigate the conformation and hyperfine structure of the nitroxide and ketyl groups in the simple H2NO and H2CO- radicals using a b initio quantum mechanical methods. The calculation of the ground-state electronic wave function, energy, and hyperfine splittings was performed using the spin-restricted LCAO-SCF method of Roothaan.7,8 A configuration interaction including all spin-adapted monoexcited configurations with three uncoupled electrons was then carried out using perturbation theor ~Under . ~ those conditions, the total energy is given by E = E,,,

i- E,,

(1)

with

In these equations, E sp and psp are the energy correction and spin density arising from spin-polarization effects respectively. The & are the doubly occupied MO’s of the ground-state, the &* are the associated virtual MO’s, and 4,, is the singly occupied orbital, responsible for the delocalization spin density Pd. Two basis sets of Gaussian-type orbitals were used. Basis I was composed of (7s,3p/3s) functions’O contracted to (3s,2p/ls). Basis I1 was built from basis I by adjunction of d orbitals on first-row atoms (C, N , 0) and p orbitals on hydrogens” (3s,2p,ld/ ls,lp). Basis I. H2NO. Starting with an approximate geometry,12 the bond lengths (NO,”) and bond angle ( H N H ) were successively varied in a cyclic procedure until a minimum in energy was located for the planar form ( N O = 1.34 8,;NH = 1.00 8,;HNH = 123’). The introduction of outof-plane geometrical parameters confirmed the stability of the planar structure. However, there is only a 0.6 kcal difference between a = 0 and 30’ (Table I). HzCO-. The initial geometry was adapted from the parent neutral moleculeI3 except for the CO bond length which has been taken as 1.34 8, by analogy with H2NO. The bond length and bond angles were varied in turn using the same cyclic procedure until the total energy was minimized. This point was achieved for a pyramidal structure (CO = 1.33 8,;CH = 1.09 8,;HCH = 117’5’; a = 24’). T h e computed inversion barrier is 0.446 kcal (156 cm-l) as shown in Table 11. Basis 11. H2NO. Starting with the above geometry, the same process of optimization led to a pyramidal equilibrium form ( N O = 1.27 8,;N H = 1.00 8,; HNH = 123’; a = 17’). The inversion barrier is then 0.064 kcal (22 cm-1).

Table 11. Total Energies of HKO- (au) Basis I

0

- 113.465440 - 113.466150

Basis I1

24 0 27

--1 13,531012 - 113,532407

-113.475202 - 113.475691 -113.544414 -113.546823

A t this stage, it is of interest to mention (Table I) that a previous unrestricted SCF c a l c ~ l a t i o nof~ ~H2NO using a noncontracted basis of a rather limited size (3s,2s,2p) led to a pyramidal equilibrium geometry ( N O = 1.32 8,;NH = 0.99 8,; HNH = 116’25’; a = 26’22’). Using the same basis set (basis 111) in the present restricted S C F formalism, we found that, all other parameters unchanged, their pyramidal form was lower than the planar one by 0.326 kcal (1 14 cm-I). Therefore, the structure found for the radical site is independent of the U H F or R H F formalism but highly sensitive to the basis. The NO bond length of 1.34 A (basis I, R H F ) or 1.32 8, (basis 111, U H F ) reduces to a value of 1.27 8, (basis 11, R H F ) in fair agreement with exp e r i m e r ~ t . This ~ . ~ suggests that polarization functions which spread far from the NO axis give a better description of the singly occupied antibonding orbital and consequently of the nitroxide group. HzCO-. Starting with the above geometry, reoptimization of bond lengths and angles has led to a quite similar pyramidal form ( C O = 1.30 8,; CH = 1.09 8,; HCH = 116’5’; a = 27’). The inversion barrier is then 0.875 kcal (306 cm-I). The preceding results are not significantly modified when adding the energetical corrections given by the first-order perturbation treatment15 to the SCF values (Tables I and 11). Introduction of polarization functions leads to a small shortening in the CO bond length, the structure found for the radical site being much less sensitive to the basis in the ketyl than in the nitroxide. In either event, the potential energy curve is very flattened around the a = 0’ position for the nitroxide and the inversion barrier, if any, is undoubtedly of the same order of magnitude as the zero point vibrational energy of the outof-plane vibration. Therefore, a precise determination of a in H z N O appears to be somewhat speculative. By contrast, the depth of the energetical well for the ketyl and the related inversion barrier are more significant, so that a pyramidal structure can be reasonably expected for H*CO-. Both radicals, in their pyramidal forms, are described by the electron configuration



(Ia‘j’(2a) ’ (3a’)2(4a‘)2(la”)2(5a’)2(6a’)2(2a”~2(7a’) (C,) It should be added that there is only one more electron in the 2a” level, essentially nonbonding a t the oxygen, compared with the lowest excited (n-a*)singlet or triplet states of formaldehyde. Interestingly, the geometrical parameters and inversion barrier determined here for H2CO- a r e in close concordance with experimental findingsi6and theoretical calculationsI7 on excited H2CO. This is in agreement with Walsh predictions on H2AB molecules1*and supports

S u b r a et al. / Conformation a n d Hyperfine Structure of H2NO a n d H3CO- Radicals

478

H,NO a.u

a.u

molecules have provided similar c o n c l ~ s i o n s .They ~ ~ also suggest that the ketyl functional group should be nonplanar. The value of the out-of-plane distorsion a t the radical site, the inversion barrier, and, of course, the a c coupling constant would be expected to depend to some extent on surrounding interactions.

I

I

0.200

H,CO

,

0200

L

/'

~

Acknowledgment. The calculations have been performed on the C D C 3600 and IBM 360-75 computers of the C.I.R.C.E. (Orsay). W e wish to thank the staff of the Computation Centre for its kind cooperation. References and Notes (1)(a) Part of the Science Thesis of Y. E., 1973;(b) Centre d'Etudes Nu-

--._

0

20 40 60

0"

o

20 40 60

ao

(G) = 1 1 5 . 3 ~(au), uc ( G ) = 401p (au); delocalization contribution (O), spin-polarization contribution (A),total spin density (+).

Figure 1. Nitrogen and carbon-13 spin densities;

UN

cleaires de Grenoble; (c) C.I.S.I.; (d) Ecole Normale Superieure de Jeunes Filles. (2)(a) General reviews may be found in A. R. Forrester. J. M. Hay, and R. H. Thomson, "Organic Chemistry of Stable Free Radicals," Academic Press, London, 1968;E. G. Rozantsev, "Free Nitroxyl Radicals," Plenum Press, New York. N. Y., 1970;A. Rassat, Pure Appl. Chem., 25, 623 (1971);(b) see, for instance, I. C. Smith, "Biological Applications of Electron Spin Resonance," Interscience, New York, N. Y., 1971;C. L. Hamilton and H. M. McConnell, "Structural Chemistry and Molecular Biology," W. H. Freeman, San Francisco, Calif., 1968. (3)(a) R. P. Shibaeva and L. 0. Atovmian, Zh. Strukt. Khim., 13, 887 (1972),(b) B. Andersen and P. Andersen, Acta Chem. Scand., 20, 2728 (1966);(c) B. Chion, A. Capiomont, and J. Lajzerowicz-Bonneteau. Acta Crystallogr., Sect. E, 28, 618 (1972);(d) J. W. Turley and F. P. Boer, Acta Crystallogr., Sect. E, 28, 1641 (1972);(e) G. Kruger and J. Boeyens, Acta Crystallogr., Sect. E, 26, 668 (1970). (4) (a) J. Lajzerowicz-Bonneteau, Acta Crystallogr., Sect. E, 24, 196 (1968);(b) 0. Bordeaux, A. Capiomont, and J. Lajzerowicz-Bonneteau, C. R. Acad. Sci., in press: (c) D. Hawley, G. Ferguson, and J. M. RobertSon, J. Chem. SOC.6, 1255 (1968);(d) A. Capiomont and J. Lajzerowicz-Bonneteau, submitted for publication; (e) A. Capiomont, B. Chion, and J. Lajzerowicz-Bonneteau, Acta Crystallogr., Sect. E, 27, 322 (1971);(f) A. Capiomont and J. Lajzerowicz-Bonneteau, Acta Crystaliogr., Sect 5, 28, 2298 (1972);(g) G. Giidewell, D. W. Rankin, A. G. Robiette, G. M. Sheldrick, and S. M. Williamson, J. Chem. SOC. A, 478

the common assumption according which n electrons have a minor influence on the molecular geometry. For the equilibrium geometries, the computed splittings (basis 11) a r e a N = +6.09 G, a 0 = -41.70 G, a H = -8.94 G for H2N0, the experimental values beingI9 l a ~ = l la~I = 11.9 G, and a c = 51.31 G, a 0 = -17.30 G, U H = -1 1.27 G for H2CO-, the experimental values being5g lac1 = 37.68 G, ( aH/ = 14.26 G. All our results are correct in sign, as compared to avail(1971). able experimental data.5d,20They are in good agreement (5) (a) M. C. R. Symons, J. Chem. Soc., 277 (1959);(b) N. Hirota and S. I. Weissman, J. Amer. Chem. Soc., 82, 4424 (1960);(c) J. E. Bennett, B. with the observed values, especially for H2CO-, if one conMile, and A. Thomas, J. Chem. SOC.A, 298 (1968);(d) G. A. Russell and siders the high sensitivity of a c to out-of-plane distorsions G. R. Underwood, J. Phys. Chem., 72, 1074 (1968);(e) W. R. Knolle and as illustrated by the variation of spin densities reported J. R. Bolton, J. Amer. Chem. Soc., 91, 541 1 (1969);(f) G. A. Russel, D. F. Lawson, H. L. Malkus, and P. R. Whittle, J. Chem. Phys., 54, 2164 below. In assessing the quality of agreement with experi(1971);(g) (G. P. Laroff and R. W. Fessenden, J. Chem. Phys., 57, 5614 ment, it should be kept in mind that a large basis set and (1972). (6) (a) J. A. Pople, D. L. Beveridge, and P. A. Dobosh, J. Amer. Soc., 90, rather extensive configuration interaction are required in 4201 (1968);(b) M. S.Gordon and J. A. Pople, J. Chem. Phys., 49, order to obtain convergence of the hyperfine s ~ l i t t i n g s , ~ ~ 4643 (1968);(c) E. T. Strom, J. Phys. Chem., 72, 4715 (1968);(d) K. especially for nitrogen.22 Morokurna, J. Amer. Chem. SOC., 91, 5412 (1969);(e) G. R. Underwood, Mol. Phys., 22, 729 (1971). Nevertheless, the comparative evolution of the nitrogen (7)C. C. J. Roothaan, Rev. Mod. Phys., 32, 179 (1960). and carbon-I 3 hyperfine splittings with a is illuminating (8)A. Veiilard, D. J. David, and P. Millie, IBMOL CDC 3600 version, Laboratoire de Chimie de I'Ecoie Normale Superieure, Paris, 1967. (see Figure 1 ) . (9)(a) P. Millie, B. Levy, and G. Berthier, Int. J. Quantum Chem., 6, 155 Both direct and indirect spin densities a r e positive. The (1972);(b) Y . Ellinger, A. Rassat, R. Subra, G. Berthier, and P. Millie, net lowering of spin polarization as cy increases is completeChem. Phys. Lett., 11, 362 (1972);(c) Y. Ellinger, A. Rassat, R. Subra, and G. Berthier, J. Amer. Chem. Soc., 95, 2372 (1973). ly overwhelmed by the delocalization contribution. The (io) E. Clementi, J. M. Andre, M. CI. Andre, D. Klint, and D. Hahn, Acta Phys., most striking feature of these curves is the higher amplitude 27, 493 (1969). (11) B. Roos and P. Siegbahn, "Polarization Functions for First and Second of the variation of a c compared with that of a\. This can Row Atoms in Gaussian-type MO-SCF Calculations," Proceedings of be simply apprehended by inspecting the distribution of spin the Seminar on Computational Problems in Quantum Chemistry, Strasbourg, 1969. populations, which a r e often identical with spin densities p" (12)J. Douady, Y . Ellinger, A. Rassat, R. Subra, and G. Berthier, Mol. Phys., in semiempirical estimations, in the equilibrium geometries 17, 217 (1969). of H2NO ( p h A = 0.24; po" = 0.76) and H2CO- (pc" = (13)(a) R. E. Lawrence and M. P. Strandberg, Phys. Rev., 83, 363 (1951); (b)G. Glockler, J. Phys. Chem., 62, 1049 (1958). 0.63; PO" = 0.33).23 The difference between these two iso(14)A. W. Saiotto and L. Burnelle, J. Chem. Phys., 53, 333 (1970). electronic compounds is that the unpaired electron is mainly (15) Ail values involving spin-polarization effects have been computed using located on the central atom in the ketyl group, contrary to quasiiocalized MO's obtained by a Foster-Boys unitary transformation of the canonical MO's: J. M. Foster and S. F. Boys, Rev. M o d Phys., the nitroxide group. This is responsible for the strong sensi32, 300 (1960). tivity of a c to geometrical changes a t the radical site. Thus (16)(a) J. C. D. Brand, J. Chem. Soc., 858 (1956);(b) W. T. Raynes. J. Chem. Phys., 44, 2755 (1966). we would anticipate an increase in a and a c with the out(17)R. Ditchfield. J. Del Bene, and J. A . Pople, J. Amer. Chem. Soc., 94, o;-plane bending. 4806 (1972). These ab initio calculations clearly show the similarity of (18)A. D. Walsh, J. Chem. Soc., 2306 (1953). (b) J. (19)(a) C. J. W. Gutch and W. A. Waters, J. Chem. Soc., 751 (1965); the isoelectronic ketyl and nitroxide radicals. They throw 0. Adams, S. W. Nicksic, and J. R . Thomas, J. Chem. Phys., 45, 654 some light on the large variety of geometries and a h found (1966). (20)Z.Luz, B. L. Silver, and C. Eden, J. Chem. Phys., 44, 4421 (1966). in nitroxides. They clearly show that the nitroxide function(21)S. Y. Chang, E. R. Davidson, and G. Vincow. J. Chem. Phys., 49, 529 al group does not possess a well-defined intrinsic geometry, (1968). as the rigid carbonyl group does, for instance. So, the con(22)R. K. Nesbet, "La Structure hyperfine magnetique des atomes et des molecules," Colloque international du CNRS, 164,1967,p 87. formation a t the radical site, the inversion barrier, and the (23)Spin populations are normalized to unity. The deviation to unity apnitrogen coupling will be critically dependent on the molecpearing for H,CO- represents the delocalization on the two hydrogens. ular f r a m e w 0 1 - k . Nitrogen ~~ inversion studies in closed shell (24)See, for example, 2,2,6,6-tetramethyI-4-piperidone-l-o~yl~~ (CNC =

Journal of the American Chemical Society

/ 97:3 / February

5, I975

479 123’5’; a = 0’; & = 15.15 G), 2,2,6,6-tetramethyl-4-piperidinol-loxy148 (CNC = 125’4’; a = 15’8’; ap, = 16.05 G), 2,2,6$-tetramethylp~peridine-l-oxyl~~ (CNC = 125’5’: a = 19’4; ap, = 16.30 G), cf. R . Briere, H. Lemaire, and A. Rassat, Bull. SOC. Chim. Fr.. 3273 (1965); 2,2,5,5-tetrarnethyl-3-pyrrolidone-l-oxyl azine3c (CNC = 112’; a = 0’;

= 14.27 G). cf. R. M. Dupeyre and A. Rassat, J. Amer. Chem. SOC., 87, 3771 (1965); 9-aza-bicyclo[3.3.l]nonan9-one-l-oxyl (CNC = 114’2‘; a = 30°1’, 6~ = 17.55 G),cf. R. M. Dupeyre and A. Rassat, J. Amer. Chem. SOC.,88, 3180 (1966). (25) J. M. Lehn, Fortschr. Chem. Forsch., 15, 31 1 (1970).

On Avoided Surface Crossings Lionel Salem,* Claude Leforestier, Gerald Segal,’ and Ross Wetmore Contributionfrom the Laboratoire de Chimie Thkoriques2Universit? de Paris-Sud, Centre d’Orsay, 91 405 Orsay, France and the Department of Chemistry, University of Southern California, Los Angeles, California 90007. Received April 16, 1974

Abstract: A classification of avoided surface crossings relevant to organic chemistry is proposed. Four different types (A, B, C, and D) are defined. An “intermediate” Hamiltonian technique is introduced for describing appropriately the avoidedcrossing region between a closed-shell surface and an open-shell surface. The gap size for avoided crossings near a symmetryallowed crossing is shown to be proportional to the transition force between the two neighboring states.

Surface crossings, whether real or intended, are playing a n increasingly important role in the interpretation of organic phenomena. It is well known that in WoodwardHoffmann thermally forbidden reactions, there is an intended crossing between the ground-state surface and the surface for the doubly-excited state.3 More recently the photochemical Norrish type I1 process has been shown to involve an intersection between the energy surface for excit. ~similar surface ed reactant and that for ground r e a ~ t a n tA crossing occurs in other major photochemical reaction^.^ Certain organic transients have two neighboring singlet surfaces with intersections and near-intersecting region^.^ Our purpose here is not to deal with rigorously allowed intersections between electronic states. The theory of these crossings is well established.6 O u r intention is to bring to light the different types of “intended” or “avoided” crossings between states of the same spin multiplicity. In certain instances the avoided crossing occurs near a real crossing. In other cases the “intention” to cross is unknown to the real surfaces; the crossings are due to an approximate starting description of the wave functions for the two neighboring surfaces. The first part of this paper contains a classification of the different types of avoided crossings which are known. In the second part of our work we introduce a method which allows for proper calculation of the neighboring surfaces in an avoided crossing region. Finally, in the third section, we demonstrate a simple law for the energy gap when a rigorously allowed crossing is destroyed concomitant to the destruction of a symmetry plane. The manner in which this gap varies with nuclear geometry is emphasized. A Classification of Avoided Surface Crossings A useful study of certain avoided crossings and of the appropriate terminology has been given by O ’ M a l l e ~The . ~ familiar distinction is made between stationary adiabatic states, which diagonalize the electronic Hamiltonian, and whose surfaces would be followed by slowly moving nuclei, and nonadiabatic (or diabatic) states for which the total electronic Hamiltonian is not diagonal. An avoided-crossing situation arises when the nonadiabatic surfaces intersect but when this intersection disappears by inclusion of the remaining off-diagonal terms

W e have been able to distinguish between four types of avoided surface crossings. The first type (A) occurs in the neighborhood of a true crossing of electronic states. The other types (B, C, D) originate from “incorrect,” nonadiabatic starting electronic wave functions, which are useful in that they are a good initial basis for discussing the “correct,” adiabatic, surfaces. In these cases a physical crossing never occurs. In the first case the previously adiabatic states and surfaces, which described correctly a physically correct intersection in one region of multidimensional space ( C , symmetry), become nonadiabatic when they are carried over into those regions of multidimensional space where the crossing is forbidden ( C l symmetry). However, they again serve as a useful starting point for the study of the new, avoiding, adiabatic states in this region of space.

Type A. Neighborhood of a Symmetry-Allowed Crossing (Destruction of the Symmetry Element). The first type of avoided crossing is encountered for molecular geometries close to, but not identical with, a symmetrical geometry in which a crossing occurs rigorously between two electronic states. A common case will be the crossing between a symmetric A’ state and an antisymmetric A” state for a molecular system with a plane of symmetry. Such interactions have been called “accidental” 6d but are quite common. Examples of such crossings in organic chemistry include ( T , K bitopic reactions4b (such as hydrogen abstraction by ketones), certain a(a,?r) tritopic reactions4b (coplanar 01 cleavage of hexadienones), etc. In the former family, one surface leads to a zwitterion, the other to a diradical; in the second family, the two surfaces lead to diradicals of different symmetry. If one or several nuclei are displaced slightly so as to destroy the symmetry plane, the crossing becomes forbidden since the adiabatic states must now have the same symmetry. Its intended character, however, shows up clearly for displacements which are not too large. Such a type A avoided crossing is illustrated in Figure 1. Salem et al.

/ On Avoided Surface Crossings